Since a content duplicate carries its file's capture time, the duplicate
check ran after the sidecar read, so a duplicate whose file already has
a capture time read and parsed a sidecar it doesn't use (more I/O on a
network share). The check now happens first and the sidecar is read for
a duplicate only when the file has no capture time.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A content duplicate whose catalog row has no capture time took the
capture time of the library photo with the same bytes, which Edit
Capture Time may have changed. The import job now resolves the file's
capture time (the file's, else its XMP sidecar's, else its modification
time) before it sets the file aside as a duplicate, and hands it over
with the file's metadata (`Duplicate::file`, a `FileFacts`). The
Lightroom import uses it; the catalog's capture time still applies on
top when the row has one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A catalog photo whose bytes match another library photo became a new record that took only the
camera description from that photo's record, so the copyright, rights, creator and other metadata
its file carries (EXIF, IPTC, embedded XMP), which a file import keeps, was lost whenever the
catalog's XMP didn't restate it. Copying them from the other record would have carried that
photo's edits instead.
The file import already reads each content duplicate before it skips it; the probed metadata now
rides on the duplicate (`Duplicate::file_meta`, not serialised) and the new record starts from it,
before the import defaults and the catalog's metadata go on top, as for a file import. Without a
probe (a path match, which this branch never sees) it keeps the exposure description only, no
longer the user-editable GPS.
Test: a content duplicate of a trashed photo, imported through the Lightroom job with a file
carrying copyright, status, usage terms and creator, keeps them and takes nothing from the other
record's edited notice, creator or GPS.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A catalog photo whose bytes match another library photo got the generic develop defaults, so a raw
with no white balance or lens setting in the catalog lost its as-shot white balance and embedded
lens corrections and rendered unlike a file import. It now gets the import defaults (camera
defaults plus the default preset), as a file import does, before the catalog's settings go on top.
Show in Library selected the lowest mapped photo id, which can be a master left in Recently Deleted
whose new virtual copy the import added; All Photos doesn't show it. It now selects the lowest
mapped photo not in Recently Deleted, and the button is left out when there is none.
Tests: a content-duplicate raw with no catalog settings equals its camera defaults (as-shot WB,
lens profile on); Show in Library skips a trashed mapped photo and offers nothing when only
trashed or vanished ids are mapped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The new record for a catalog photo whose bytes match another library photo copied that whole
record and reset a few fields, so the other photo's edits, history, versions, ratings and written
metadata (title, keywords, face regions) carried over, from a trashed photo too. It now starts as a
new photo and takes only what the bytes determine: format, size, kind, capture time, camera EXIF,
GPS, focus point, content hash, embedded lens data and preview-only state. The catalog's own
settings and metadata apply on top as before.
Tests: the duplicate test gives the other photo edits, history, a version, a title, keywords and a
Local baseline and checks none reach the new record; a reimport after a master and its virtual copy
were trashed counts each once.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A catalog photo whose file matched a photo in Recently Deleted by content was cloned with the trashed
record's `deleted` flag (and its virtual-copy fields), so it landed in Recently Deleted while the report
counted it as imported. A new virtual copy of a trashed master did the same. Both are now live records,
as Create Virtual Copy already does.
Catalog photos matched by path to a trashed photo stay there (the Lightroom import's OnDeleted::Skip,
#706); the report counts them as `inRecentlyDeleted` and the result window says so in one line.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- RawImage::develop demosaics only the default crop with AHD when there
is no OpcodeList3 (ahd_window), instead of demosaicing the margin too
and then copying the crop out of the full picture on one thread.
- Gpu::upload writes through queue.write_buffer_with and copies into
the staging buffer in parallel parts. create_buffer_init zeroed a
staging buffer and then copied over it, two single-threaded passes
over a full-size photo's ~290 MB.
- finish_and_read copies the mapped readback out in parallel parts (one
thread read the 91 MB of a 24 MP export at ~2 GB/s, ~45 ms).
Output is byte-identical (ahd_window_is_the_crop_of_the_whole,
uploads_read_back_exactly).
Export of a 24 MP ARW to full-size JPEG (examples/export_bench, 75
interleaved runs each on a shared 16-thread Ryzen):
main 802 ms median / 614 ms min -> this branch 647 / 525 (-19.3% / -14.5%).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- finish_8bit/finish_deep hand back the render itself when there is no
output sharpening and no watermark. Before, every export cloned the
image (~100 MB at 24 MP) and output_sharpen built a ~290 MB float copy
of it even for SharpenFor::None, all on one thread.
- Output sharpening's box blurs run row-parallel (par_rows); each pixel
is computed as before.
- The parallel JPEG encoder converts each band's rows to YCbCr inside the
band's task instead of first filling whole-image planes (~290 MB for
4:4:4 at 24 MP, written once and read back).
- fdct_quant rounds with plain arithmetic (round_i16) instead of
f32::round, a libm call per coefficient without SSE4.1; it equals
`x.round() as i16` for all 2^32 inputs.
Output is byte-identical (pinned by encoded_bytes_are_pinned and
sharpened_pixels_are_pinned, whose hashes come from main).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
examples/export_bench imports the files given and runs export_photo
(decode, full-size render, encode) N times, forgetting the decoded
original between runs like a batch export does, and prints the median
and minimum wall time. OPT takes export options as app.export JSON,
PHASES=1 times render and encode separately, OUT=dir writes the files
for byte comparisons between builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Use craft-ui's dropdown so menus grow back after shrinking
The album picker keeps its choices and search field in a fixed header and scrolls only
the list; combo_box and the print dialog's combos use craft_ui::ComboBox. egui::ComboBox
is now disallowed by clippy.
* chore(deps): craft-libs 60d60d48 (dropdown fixes)
* chore(deps): craft-libs 6af63481 (dropdown fixes)
* chore(deps): craft-libs 4d7b82f7 (dropdown fixes)
* chore(deps): craft-libs 92d392cf (#387 merged)
Reuse cached folder listings through Arc and allow tests to inject Local places.
Show browsed folders below hidden ancestors as session roots. Apply the same
hidden-entry checks to folder walks, browsing, import scanning, and Local.
Return an error when either file index is zero. Keep the resolved-path fallback for lenient comparisons. Test both unknown-identity orderings and distinguish identity lookup failures from known-original refusals.
Merge main at 662df66f.
